A statistical analysis of learning experiments involving an interactive human-machine system

This thesis is a statistical analysis of three learning experiments conducted in the Synnoetics Laboratory at the University of Wisconsin with engineering and vocational students. A comparison is.made of the effectiveness of three instruction controls: pre-programmed instruction under machine control, learner control in which students specifically requested subject matter, and collaborative or guided learning which permitted students flexibility in order and amount of subject matter received. Wiien students had acquired the fundamentals, each of the two subsets of instruction control Vx=s amplified by one of two methods of feedback for students1 problem solving. One method was an incremental schedule ir. x-rhich machine response occurred after attempted partial solutions to problems while an integral schedule reserved machine response until student had attempted a complete solution. The data analyzed consists of scores from post, retention, and standardized tests, and measures of students * intercommunications with the machine and its associated subject-matter base. Students directing their own learning and solving problems.integrally had the lowest rate of processing information while those solving problems incrementally are observed to have high Effective Learning Capacity and high Effective Learning Rate. Students who directed their own learning are observed to make fewest subject-matter accesses once they have begun to problem solve.Computer Science, Department o